Images

Image editing

Edit or extend an existing image from a prompt and an optional mask.

POST /v1/images/edits

Creates an edited or extended image from one or more source images and a prompt. The endpoint follows the OpenAI Images API and accepts multipart/form-data, so the official OpenAI SDKs work unchanged with base_url set to https://rutoken.ai/v1. Supported models include gpt-image-1 and dall-e-2, depending on what is enabled in the Model Square.

Authentication

Send your key in the Authorization: Bearer sk-xxxx header.

Request body

Content type: multipart/form-data.

Parameter	Type	Required	Description
`image`	file	Yes	Image to edit. `gpt-image-1`: PNG, WEBP or JPG under 50 MB; repeat the field (`image[]`) to send up to 16 images. `dall-e-2`: one square PNG under 4 MB.
`prompt`	string	Yes	Description of the desired result. Max 32000 characters for `gpt-image-1`, 1000 for `dall-e-2`.
`model`	string	Yes	Model ID, e.g. `gpt-image-1`.
`mask`	file	No	PNG whose fully transparent areas mark where the image should be edited. Must have the same size as `image`.
`n`	integer	No	Number of images, 1–10. Default `1`.
`size`	string	No	`gpt-image-1`: `1024x1024`, `1536x1024`, `1024x1536` or `auto`. `dall-e-2`: `256x256`, `512x512`, `1024x1024`.
`quality`	string	No	`gpt-image-1`: `low`, `medium`, `high` or `auto`.
`response_format`	string	No	`url` or `b64_json` (`dall-e-2` only; `gpt-image-1` always returns `b64_json`).
`background`	string	No	`gpt-image-1` only: `transparent`, `opaque` or `auto`.
`output_format`	string	No	`gpt-image-1` only: `png`, `jpeg` or `webp`.
`input_fidelity`	string	No	`gpt-image-1` only: `high` or `low` — how closely to preserve input details such as faces.
`stream`	boolean	No	Stream partial images, when the upstream supports it.
`user`	string	No	End-user identifier.

Request example

cURL

curl https://rutoken.ai/v1/images/edits \
  -H "Authorization: Bearer $API_KEY" \
  -F model="gpt-image-1" \
  -F image="@room.png" \
  -F mask="@mask.png" \
  -F prompt="Add a large green plant next to the sofa" \
  -F size="1024x1024"

Python

import base64
from openai import OpenAI

client = OpenAI(base_url="https://rutoken.ai/v1", api_key="sk-xxxx")
resp = client.images.edit(
    model="gpt-image-1",
    image=open("room.png", "rb"),
    mask=open("mask.png", "rb"),
    prompt="Add a large green plant next to the sofa",
    size="1024x1024",
)
with open("room-edited.png", "wb") as f:
    f.write(base64.b64decode(resp.data[0].b64_json))

Node.js

import fs from "node:fs";
import OpenAI from "openai";

const client = new OpenAI({ baseURL: "https://rutoken.ai/v1", apiKey: "sk-xxxx" });
const resp = await client.images.edit({
  model: "gpt-image-1",
  image: fs.createReadStream("room.png"),
  mask: fs.createReadStream("mask.png"),
  prompt: "Add a large green plant next to the sofa",
  size: "1024x1024",
});
fs.writeFileSync("room-edited.png", Buffer.from(resp.data[0].b64_json, "base64"));

Response

json

{
  "created": 1760000000,
  "data": [
    { "b64_json": "iVBORw0KGgoAAAANSUhEUgAA..." }
  ],
  "usage": {
    "total_tokens": 1650,
    "input_tokens": 450,
    "output_tokens": 1200,
    "input_tokens_details": { "text_tokens": 30, "image_tokens": 420 }
  }
}

Field	Type	Description
`created`	integer	Unix timestamp.
`data[].b64_json`	string	Base64-encoded result image.
`data[].url`	string	Result URL, when `response_format` is `url`.
`usage`	object	Token usage for token-billed models such as `gpt-image-1`.

Notes

The gateway also accepts a JSON body for this endpoint and forwards it upstream as is; use it only if your upstream model documents a JSON form of image editing.
POST /v1/images/variations is not implemented.